Predicting postmortem interval based on microbial community sequences and machine learning algorithms.

Area of Science:

• Forensic Microbiology
• Decomposition Biology
• Computational Biology

Background:

• Microbial succession and behavior during decomposition are poorly understood.
• Predictable microbial changes after death can aid in estimating the postmortem interval (PMI).
• Accurate PMI estimation is critical in forensic investigations but presents challenges.

Purpose of the Study:

• To investigate microbial succession patterns during corpse decomposition.
• To estimate postmortem interval (PMI) using microbial data and machine learning.
• To identify key microbial species indicative of decomposition stages.

Main Methods:

• Microbial community characterization and microbiome sequencing from mouse organs (brain, heart, cecum).
• Application of machine learning algorithms: random forest (RF), support vector machine (SVM), and artificial neural network (ANN).
• Analysis of microbial data to identify informative species and build predictive models for PMI.

Main Results:

• Significant differences in microbial communities were observed between death and advanced decay stages.
• Enterococcus faecalis, Anaerosalibacter bizertensis, and Lactobacillus reuteri were identified as key species.
• The ANN model using cecal microbiome data achieved a mean absolute error of 1.5 ± 0.8 hours (24h) and 14.5 ± 4.4 hours (15 days).

Conclusions:

• Microbial succession patterns are reliable indicators of decomposition stages.
• Machine learning models, particularly ANN with cecal data, offer accurate PMI estimation.
• This integrated approach provides a robust tool for forensic science in determining PMI.